Description
ChatCompletions API specific clients can be configured with Responses API endpoints and no error or warning is generated. The configuration seems to work but behind closed doors the two just don't get along. Coding agents I've assigned to MAF builds get mixed up on the endpoint to client mappings and 9 times out of 10 wire up Responses API endpoints to chat clients.
Also, the framework has a naming collision between ChatClient (OpenAI SDK, Chat-Completions wire) and ChatClientAgent (MAF, transport-agnostic) that confuses the heck out of coding agents - and me too. Can we depreciate ChatClientAgent in favor of ModelClientAgent? It would be great if all endpoint agnostic objects refrained from naming themselves with endpoint specific terms.
Great product, keep up the outstanding work!
Code Sample
Error Messages / Stack Traces
Package Versions
Latest
.NET Version
10
Additional Context
No response
Description
ChatCompletions API specific clients can be configured with Responses API endpoints and no error or warning is generated. The configuration seems to work but behind closed doors the two just don't get along. Coding agents I've assigned to MAF builds get mixed up on the endpoint to client mappings and 9 times out of 10 wire up Responses API endpoints to chat clients.
Also, the framework has a naming collision between ChatClient (OpenAI SDK, Chat-Completions wire) and ChatClientAgent (MAF, transport-agnostic) that confuses the heck out of coding agents - and me too. Can we depreciate ChatClientAgent in favor of ModelClientAgent? It would be great if all endpoint agnostic objects refrained from naming themselves with endpoint specific terms.
Great product, keep up the outstanding work!
Code Sample
Error Messages / Stack Traces
Package Versions
Latest
.NET Version
10
Additional Context
No response